Objective Elbow flexion contractures are common complications of neonatal brachial plexus palsy, but evidence on how to treat these contractures is weak. This study compared the treatment of elbow flexion contractures using a dynamic orthosis or serial circular casting. Methods A randomized controlled trial was conducted with one-year follow-up. Children with an elbow flexion contracture of ≥30° were treated with either a night-worn dynamic orthosis for one year or serial casting for four weeks followed by night splinting. For pragmatic reasons, some participants were included in an open part of this study, this group was also analyzed separately. Degree of contracture and goal attainment scaling was evaluated at baseline and after 8, 20 and 54 weeks. Results 55 patients were analyzed in this trial, 32 of whom were randomized to treatment. At one-year follow-up of the randomized group, both dynamic splinting (median −8.5°, interquartile range [IQR] −13.5, −5) and serial casting (median −11.0°, IQR −16, −5) resulted in significant reduction of contracture ( P < 0 .001). The reduction was significantly greater with serial casting in the first 20 weeks, but not at one-year follow-up ( P = 0.683). In the entire cohort, the individual functional goals had been reached in 24 out of 32 cases (80%) of dynamic splinting and 18 out of 23 cases (82%) of serial casting, respectively. Conclusion The dynamic night orthosis is comparable to serial casting for treating elbow flexion contractures in children with brachial plexus birth injury. We recommend selecting one of these treatment modalities in close consultation with parents and patients.